ARTICLE

Received 20 Sep 2013 | Accepted 16 Oct 2013 | Published 18 Nov 2013 DOI: 10.1038/ncomms3788 MEIOB exhibits single-stranded DNA-binding and exonuclease activities and is essential for meiotic recombination

Mengcheng Luo1, Fang Yang1, N. Adrian Leu1, Jessica Landaiche2, Mary Ann Handel3, Ricardo Benavente4, Sophie La Salle2 & P. Jeremy Wang1

Meiotic recombination enables the reciprocal exchange of genetic material between parental homologous chromosomes, and ensures faithful chromosome segregation during meiosis in sexually reproducing organisms. This process relies on the complex interaction of DNA repair factors and many steps remain poorly understood in mammals. Here we report the identi- fication of MEIOB, a meiosis-specific protein, in a proteomics screen for novel meiotic chromatin-associated proteins in mice. MEIOB contains an OB domain with homology to one of the RPA1 OB folds. MEIOB binds to single-stranded DNA and exhibits 30–50 exonuclease activity. MEIOB forms a complex with RPA and with SPATA22, and these three proteins co-localize in foci that are associated with meiotic chromosomes. Strikingly, chromatin localization and stability of MEIOB depends on SPATA22 and vice versa. Meiob-null mouse mutants exhibit a failure in meiosis and sterility in both sexes. Our results suggest that MEIOB is required for meiotic recombination and chromosomal synapsis.

1 Center for Animal Transgenesis and Germ Cell Research, Department of Animal Biology, University of Pennsylvania School of Veterinary Medicine, 3800 Spruce Street, 200E Old Vet, Philadelphia, Pennsylvania 19104, USA. 2 Department of Biochemistry, Chicago College of Osteopathic Medicine, Midwestern University, Downers Grove, Illinois 60515, USA. 3 The Jackson Laboratory, 600 Main Street, Bar Harbor, Maine 04609, USA. 4 Department of Cell and Developmental Biology, Biocenter of the University of Wu¨rzburg, Am Hubland, Wu¨rzburg 97074, Germany. Correspondence and requests for materials should be addressed to P.J.W. (email: [email protected]).

NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications 1 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788

eiosis, the hallmark of gametogenesis in sexually cells14, and from XXY* testes, which lack all germ cells and thus reproducing organisms, creates haploid germ cells from contain only somatic cells15. By subtracting these two data sets Mdiploid progenitors and ensures their genetic diversity from the P18 data set, with a stringent cutoff value of at least 3 through recombination. Physical links between homologous unique peptides in P18 but none in P6 and XXY*, we identified 51 chromosomes permit the formation of crossovers required for putative meiotic chromatin-associated proteins (Fig. 1d and allele recombination and also facilitate accurate chromosome Supplementary Table S1). segregation at the first meiotic cell division1,2. Meiotic Identification of previously known meiotic chromatin-asso- recombination is initiated at DNA double-strand breaks (DSBs) ciated proteins confirmed the validity of our approach: 13 of generated by SPO11 (ref. 3); these DSBs become resected from 51 proteins identified are known to localize to meiotic the 50-end, yielding extended 30-single-stranded DNA (ssDNA) chromatin (SYCP1, SYCP2, SYCP3, SYCE2, SMC1B, HOR- overhangs4. Replication protein A (RPA) coats ssDNA, forming MAD1, H2AX, MDC1, TERF1, TDP1, PIWIL1, PIWIL2 and nucleoprotein filaments that prevent degradation and secondary YBX2; Supplementary Table S1), including all three synaptone- structure formation. Subsequent invasion of the 30-single strand mal complex proteins. Furthermore, the genes encoding 19 into the duplex of the homologue is mediated by DMC1 and proteins identified in our screen had been subjected to targeted RAD51. Strand invasion causes the displacement of one strand deletion in mice: 11 mutants exhibited meiotic arrest, 2 showed of the homologue, resulting in D-loop formation5. The invading post-meiotic spermiogenic arrest, 4 displayed embryonic lethality 30-end primes new DNA synthesis, such that D-loop extension (with an unknown role in meiosis) and 2 were dispensable for towards the second end allows for the capture of the 30-ssDNA fertility. The identification of these known meiosis factors end of the other break (second end). End ligation leads to the validates our biochemical/proteomic approach and suggests formation of double Holliday junctions. Although early steps of that the other 32 proteins identified in our screen are likely to meiotic recombination (DSB formation, end resection, strand be novel meiotic chromatin-associated proteins (Supplementary invasion and D-loop formation) have been relatively well Table S1). characterized, the intermediate steps of this process, such as second-end capture, are poorly defined. MEIOB is a novel meiosis-specific protein. To test whether the RPA is a conserved ssDNA-binding heterotrimer complex of newly identified putative meiotic chromatin-associated proteins RPA1, RPA2 and RPA3 (ref. 6). The four oligonucleotide-binding indeed localize to meiotic chromatin and function in meiosis, we (OB) folds of RPA1 mediate ssDNA binding of the RPA focused on one of these uncharacterized proteins—MEIOB. complex7. RPA is essential for many aspects of DNA MEIOB (470 amino acids (aa)) is predicted to contain an OB fold metabolism, notably DNA replication, DNA repair and meiotic (aa 167–272) with limited homology to the ssDNA-binding recombination. During meiotic recombination, RPA binds to domain-B OB domain of RPA1 (Supplementary Fig. S1). MEIOB various ssDNA intermediates, including resected 30-ssDNA and homologues are present in mammals, chicken, fugu fish, Droso- the D-loop. Studies of mutant alleles of yeast RPA1 have phila and Arabidopsis. Western blot analyses using specific anti- demonstrated that RPA is required for meiotic recombination8. bodies raised against MEIOB confirmed its association with RPA is indispensible for mouse development; a homozygous chromatin; MEIOB was also detectable in the cytoplasmic and point mutation in Rpa1 causes early embryonic lethality9. soluble nuclear fractions of the testis (Fig. 1e). Although abun- Here we report the identification of meiosis-specific with OB dantly expressed in the testis, MEIOB was not detectable in domains (MEIOB), a meiosis-specific protein with homology to somatic tissues when examined by western blotting (Fig. 1f). one OB fold of RPA1, in a systematic proteomics screen for MEIOB localized predominantly to the nuclei of spermatocytes in meiotic chromatin-associated proteins. We find that mouse the testis (Supplementary Fig. S2). Therefore, MEIOB is a novel MEIOB has evolved to fulfill an essential function in meiosis. meiotic chromatin-associated protein.

Results MEIOB co-localizes with RPA in foci on meiotic chromosomes. Proteomics screen for meiotic chromatin-associated proteins. Immunostaining of spread nuclei of spermatocytes showed that To systematically identify proteins that are associated with MEIOB formed discrete foci on chromosomes, foci that exhibited meiotic chromosomes, we isolated chromatin from postnatal day a dynamic distribution pattern during meiosis (Fig. 2a–j). MEIOB 18 (P18) testes, which lack post-meiotic germ cells and are thus foci were absent at the early leptotene stage (Fig. 2a), became first highly enriched for meiotic germ cells. Our adaptation of a detectable on late leptotene chromosomes (Fig. 2b), formed arrays chromatin purification protocol for testis samples yielded highly along synaptonemal complexes in zygotene and early pachytene pure chromatin (Fig. 1a)10,11. SYCP2, a synaptonemal complex spermatocytes (Fig. 2c,d), and disappeared in late pachytene protein associated with meiotic chromatin12, was present in the spermatocytes (Fig. 2e). At a range of 100–200 per cell, MEIOB chromatin fraction, whereas the cytoplasmic germ cell-specific foci peaked at the zygotene stage (Fig. 2k). Notably, MEIOB foci protein TEX19 was not detectable in the chromatin fraction by were formed along the partially synapsed sex chromosomes at the western blot analysis (Fig. 1b)13. To improve detection of proteins pachytene stage (Fig. 2d). In addition, MEIOB formed foci on with low abundance, we separated chromatin-associated proteins meiotic chromosomes in fetal oocytes (Supplementary Fig. S3), from P18 testes by SDS–polyacrylamide gel electrophoresis implying functional relevance of MEIOB for meiosis in both (PAGE) and analysed ten different size fractions by tandem sexes. MEIOB was absent from the adult ovary (Fig. 1f), in which mass spectrometry (MS/MS; Fig. 1c). A total of 1,300 proteins oocytes are at the diplotene stage. No focal staining was observed with at least one unique peptide were identified. in Meiob À / À spermatocytes with the antibody used for these In addition to meiotic cells, P18 testes contain mitotic germ analyses, demonstrating that the antibody is specific for MEIOB cells and somatic cells (Fig. 1d). To identify proteins only and does not crossreact with RPA1 (Supplementary Fig. S4a). associated with meiotic chromatin, we applied a subtractive We next evaluated the potential interaction of MEIOB with approach, excluding chromatin-associated proteins present in other meiotic recombination-related proteins and found that mitotic cells. Using the same biochemical/proteomics protocol, MEIOB co-localized with RPA, which forms foci on prophase I we identified chromatin-associated proteins from postnatal day 6 meiotic chromosomes16–18. Immunofluorescence analyses with (P6) mouse testes, which contain spermatogonia but no meiotic antibodies specific for both the RPA1 and RPA2 subunits of RPA

2 NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788 ARTICLE

100 mg P18 testes Homogenize in Buffer A 1D P18 testes PAGE 1 slice/tube 300 g, 5 min Sup (cytoplasmic extract) Pellet (nuclei) Cut into Each 10 slices slice ChromatinNuc. membraneCytoplasm Homogenize in Buffer B MS/MS 250 100 g, 30 s SYCP2 Pellet (nuc. membrane) 150 50 Supernatant TEX19 (chromatin + nuclear extract) Histone 15 1.7 M sucrose cushion Sup (nuclear extract, H3 50,000 g, 1 h and so on) 75 μg Pellet (P18) (purified chromatin)

P18 testis

(mitotic + meiotic germ cells) Total CytoplasmicNuclear extractChromatin extract MCAPs MEIOB 50 ≥ 51 ( 3 unique peptides) 50 TEX19 Histone H3 15

P6 testis XXY* testis

(mitotic germ cells) (no germ cells) Kidney Lung Brain Heart Spleen Sk muscle Testis Ovary MEIOB 50 50 ACTB 37 Figure 1 | Proteomic identification of meiotic chromatin-associated proteins from mouse testes. (a) Biochemical purification of chromatin from P18 testes. (b) Purity assessment of isolated meiotic chromatin by western blotting. SYCP2 is known to associate with meiotic chromatin. TEX19 is a cytoplasmic protein control. Histone H3 is a ubiquitous chromatin-associated protein. (c) Identification of chromatin-associated proteins by MS/MS. The gel was cut into ten slices, separating bands with abundant proteins from those with lower protein content. (d) Systematic identification of meiotic chromatin-associated proteins (MCAPs) by subtractive analysis of three proteomic data sets. Proteins represented with at least three unique peptides in P18 testes, but absent from both P6 and XXY* testes, were considered candidate meiotic chromatin-associated proteins. (e) Association of MEIOB with chromatin in P18 testes. Western blot analysis was performed using 20 mg protein from cytoplasmic extract, nuclear extract and chromatin. TEX19 and histone H3 serve as cytoplasmic and chromatin controls, respectively. (f) MEIOB protein expression in adult mouse tissues. ACTB serves as a loading control. Note that heart and skeletal muscle contain little ACTB. Protein molecular mass standards are shown in kDa. confirmed complete focal association with MEIOB (Fig. 2l–u). As MEIOB binds to ssDNA. A central OB domain is the only previously shown, RPA foci gradually replace RAD51 foci during predicted protein motif in MEIOB (Fig. 3a). OB folds are compact meiotic recombination; thus, only a subset of RPA foci overlap barrel-like structures that can recognize and interact with single- with RAD51 foci17. The relative distribution pattern of MEIOB stranded nucleic acids or those with an unusual structure21,22. and RAD51 foci is similar to that of RPA and RAD51 foci. Early We expressed the central region of MEIOB including the OB fold leptotene spermatocytes contained a large number of RAD51 foci (aa 136–307; Fig. 3a, construct A) as a glutathione S-transferase but lacked MEIOB foci (Supplementary Fig. S5a). In zygotene (GST) fusion protein and assessed its nucleic acid binding and pachytene spermatocytes (Supplementary Fig. S5b–g), the capacity. Truncated MEIOB (construct A) bound to ssDNA with distribution of MEIOB and RAD51 foci appeared to be dynamic: an apparent Kd of 0.4 mM (Fig. 3b,c). Truncated MEIOB exhibited some foci contained both proteins, some contained RAD51 only high affinity for ssDNA at least 24 nucleotides (nt) in length and and others contained MEIOB only. RAD51 and DMC1 always lower affinity for shorter fragments (Fig. 3d). Truncated MEIOB co-localize in foci on meiotic chromosomes19. As expected, also bound to dsDNA with 50-or30-single-stranded flaps, but not MEIOB co-localized with DMC1 in only a subset of foci to dsDNA with blunt ends (Fig. 3e). These experiments (Supplementary Fig. S6a–d0). In a subset of foci, MEIOB also demonstrate that MEIOB is an ssDNA-binding protein. co-localized with MSH4, a member of the ZMM group involved To identify essential elements of the ssDNA-binding domain in meiotic recombination (Supplementary Fig. S6e–h0)20. Thus, of MEIOB, we expressed protein segments with additional MEIOB and RPA always co-localize, whereas MEIOB co-localizes amino- and carboxy-terminal truncations and point mutations with RAD51, DMC1 or MSH4 only in a subset of foci. (constructs B–G; Fig. 3a,f). We found that deletion of aa 277–307, Ultrastructural analysis by immunogold electron microscopy located C-terminal to the OB fold, resulted in a tenfold reduction showed MEIOB signal in the electron-dense nodule between in the binding affinity (Kd for construct B: 3.9 mM), whereas lateral elements of the synaptonemal complex (Fig. 2v). These deletion of aa 136–178 (construct C) abolished DNA-binding results suggest that MEIOB associates with RPA during meiotic activity. On the basis of the crystal structure of human RPA1 recombination. (ref.23), several residues in the ssDNA-binding domain-B OB

NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications 3 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788

MEIOB MEIOB MEIOB MEIOB MEIOB SYCP2 SYCP2 SYCP2 SYCP2 SYCP2 XY

XY

Early leptotene Late leptotene Zygotene Early pachytene Late pachytene

MEIOB MEIOB MEIOB MEIOB MEIOB

250 MEIOB RPA1 MEIOB RPA1 200

150 MEIOB RPA1 100

50 MEIOB foci per cell 0 MEIOB RPA1 l-Le Zy e-Pa (n=10) (n=25) (n=22) MEIOB RPA2 MEIOB RPA2

LE LE MEIOB RPA2

MEIOB RPA2

Figure 2 | MEIOB co-localizes with RPA in foci on meiotic chromosomes. (a–e) Distribution pattern of MEIOB foci on chromatin of spermatocytes from early leptotene through late pachytene stages. (f,j) the same cells in a–e are shown with MEIOB labelling only. (k) Number of MEIOB foci in spermatocytes at successive stages of meiotic prophase I. Stages were determined by SYCP2 staining. l-Le, late leptotene; Zy, zygotene; e-Pa, early pachytene; n, number of cells counted. The line indicates the average. (l–p) MEIOB co-localizes with RPA1 on the meiotic chromosomes of spermatocytes. The same spermatocyte is shown with MEIOB labelling only (l), RPA1 labelling only (m) and in a merged image (n). Enlarged views of the marked chromosome in n are shown in o and p (without or with offset channels). (q–u) MEIOB co-localizes with RPA2 on the meiotic chromosomes of spermatocytes. The same spermatocyte is shown with MEIOB labelling only (q), RPA2 labelling only (r) and in a merged image (s). Enlarged views of the marked chromosome in s are shown in t and u (without or with offset channels). (v) Electron microscopy with immunogold (6 nm particles, arrowhead)- labelled MEIOB reveals signal between the lateral elements (LE) in spermatocytes. Scale bars, 10 mm(a–s); 0.2 mm(v). domain are involved in DNA binding; these are conserved in the 30-exonuclease activity. Truncated MEIOB did not process 50-flap MEIOB OB domain (Supplementary Fig. S1). Analysis of mutant ssDNA, showing that it lacks 50-exonuclease or endonuclease protein segments with point mutations at such residues identified activity (Fig. 4d). Rather, truncated MEIOB cleaved nucleotides R235 of MEIOB to be essential for DNA binding. Residue S251 from the 30-end of 30-radiolabelled oligos (Fig. 4e), demonstrating was important but not required for DNA binding, as the S251A that it is a 30-exonuclease. The 30-exonuclease activity of trun- mutation reduced the binding affinity by 4.5-fold (Kd for cated MEIOB was robust for ssDNA of Z18 nt and lower for construct F: 1.8 mM). shorter oligonucleotides (5–15 nt; Fig. 4e), suggesting that the enzymatic activity of truncated MEIOB increases with stronger DNA binding. The ability of truncated MEIOB to digest the MEIOB exhibits ssDNA-specific 30–50 exonuclease activity. 30-flap DNA further validated its 30-exonuclease activity (Fig. 4f). When the in vitro binding assays with 50-32P-labelled ssDNA The exonuclease activity of truncated MEIOB on the 30-flap were performed in the presence of Mg2 þ , a cleaved product was similar to that on ssDNA. Truncated MEIOB digested appeared proportionally to MEIOB concentration, revealing that 30-flaps efficiently to B8 nt, digested 6- and 8-nt 30-flaps very 2 þ 0 truncated MEIOB (construct A in Fig. 3a) exhibits Mg - inefficiently, but had no effect on 4-nt 3 -flaps (Fig. 4g). At 0.8 mM, dependent nuclease activity (Fig. 4a). Truncated MEIOB cleaved truncated MEIOB digested ssDNA efficiently only to a length of ssDNA into oligonucleotides of six to predominantly eight bases 8 nt after 10 min of incubation (Fig. 4h). With a tenfold excess in length (Fig. 4b). Although its binding affinity to oligonucleo- (8 mM), truncated MEIOB further digested ssDNA to o5ntor tides of r18 nt was low (Fig. 3d), truncated MEIOB cleaved even completely (Fig. 4h). In conclusion, truncated MEIOB ssDNA fragments as short as 10–15 nt into 8-nt oligonucleotides exhibits 30-exonuclease activity specific for ssDNA in vitro. (Fig. 4c). The generation of 50-labelled 6- to 8-nt products However, we cannot exclude the possibility that in vitro bio- indicated that truncated MEIOB has either endonuclease or chemical analyses on truncated MEIOB proteins may not reflect

4 NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788 ARTICLE

OB fold 1 MEIOB 1 470 MEIOB (mM) 0.8 Construct 167 272 ssDNA binding A 136 307 +++ 0.6 GST (4 mM) 0.4 0.6 0.8 1 4 0 0.1 0.2 2 B 136 276 + 0.4 276 – C 179 * Fraction 307 – Bound 0.2

D 179 ssDNA bound E R235A 136 307 – * * 0 F S251A 136 * 307 ++ Free ssDNA 02341 36 nt G R235A S251A 136 * * 307 – MEIOB (mM)

MEIOB (0.8 mM) MEIOB construct (1 mM) -flap -flap ′ ssDNA (nt) ′ 15 18 22 24 26 30 36 20 28 ds (36 bp) 5 3

* * GST ABCDEFG MEIOB * * +++ * Bound (0.8 mM) Bound Bound

Free *Free *Free 36 nt Free kDa (mM) 0.4 3.9 1.8 % Bound 0 0 1 3 48 19 69 82 90 % Bound 0 23 33

Figure 3 | MEIOB binds to ssDNA but not dsDNA. (a) MEIOB contains an OB fold. All MEIOB constructs were expressed as GST fusion proteins. Only construct A (MEIOB central region; aa 136–307) was used in experiments illustrated in b through e. The summary of ssDNA-binding activity is based on the results from panel f: þþþ, strong; þþ, intermediate; þ , weak; À , no binding. The Kd values are shown in f.(b) Truncated MEIOB binds to ssDNA. The ssDNA (d36GT) was 50-labelled with 32P (indicated by asterisk). (c) Binding curve for truncated MEIOB-ssDNA (d36GT) from triplicate electrophoretic mobility shift assay (EMSA) experiments (mean±s.d.). (d) Binding of truncated MEIOB to ssDNA oligonucleotides of various lengths. All oligonucleotides [d(GT)n or d(GT)nG] were 50-32P-labelled. (e) Truncated MEIOB binds to 50- and 30-ssDNA flap (30 nt) but not dsDNA. One strand of the dsDNA and the 50-flap DNA were 50-32P-labelled. The long strand in the 30-flap DNA was 30-32P-labelled. (f) EMSA analysis of various MEIOB constructs depicted in a. GST was used as negative control. precisely the activity of the full-length protein in vitro, in vivo,or and in adults (Supplementary Fig. S7). TdT-mediated dUTP nick in partner with other proteins. end labeling (TUNEL) assays showed that Meiob-deficient We next tested the nuclease activity of various MEIOB mutants spermatocytes underwent apoptosis (Fig. 5g), suggesting that with truncations and point mutations (Fig. 3a). We found that all mutant spermatocytes are eliminated at the pachytene the mutant proteins exhibited much lower nuclease activities than checkpoint24. construct A (Fig. 4i). Interestingly, similar to construct A, Ovaries from adult Meiob À / À female mice were small and constructs B and F digested the 36 nt ssDNA to 8 nt. In contrast, devoid of oocytes of any developmental stage (Fig. 5h). To constructs C, D, E and G produced digested oligonucleotides determine the time point of oocyte loss, we examined ovaries at longer than 8 nt, suggesting lower enzymatic processivity. As different ages, using anti-YBX2 antibodies to identify oocytes25. constructs A, B and F bind to ssDNA, but C, D, E and G do not Meiob À / À ovaries contained an abundant number of oocytes at (Fig. 3f), these data suggest that the DNA-binding property of birth (P0, Fig. 5i), but were almost devoid of oocytes by postnatal MEIOB might stimulate its enzymatic processivity. day 1 (Fig. 5j), with complete lack of oocytes by day 2 (Fig. 5k). TUNEL assays revealed that oocytes in Meiob À / À ovaries underwent massive apoptosis at postnatal day 1 (Fig. 5l), which MEIOB is required for meiosis and fertility in both sexes.To is in agreement with a complete oocyte loss within the first 2 days uncover the requirement of MEIOB for meiosis, we disrupted the after birth. Complete early postnatal loss of oocytes is Meiob gene (Fig. 5a). Meiob À / À mice were viable and appeared to characteristic of recombination-defective (Dmc1, Msh5 or Atm) be healthy. Western blotting demonstrated the absence of MEIOB mouse mutants26. Therefore, MEIOB is essential for meiosis in protein in Meiob À / À testes (Fig. 5b), confirming that both sexes. the Meiob mutant was null. Interbreeding of heterozygous (Meiob þ / À ) mice yielded a normal Mendelian ratio (182:357:169) of Meiob þ / þ , Meiob þ / À ,andMeiob À / À offspring. MEIOB is required for chromosomal synapsis. To elucidate the Both Meiob À / À males and females were sterile. Disruption of molecular defects of Meiob-deficient meiotic germ cells, we Meiob resulted in dramatically reduced testis size (Fig. 5c). Testes monitored the process of chromosomal synapsis by assessing the from 8-week-old Meiob À / À males weighed 70% less than wild distribution of proteins of the central (SYCP1) and axial/lateral type (54.5±6.7 mg versus 184.0±10.5 mg per pair; n ¼ 4; elements (SYCP2) of synaptonemal complexes (Supplementary Student’s t-test, Po0.0001). Histological analysis revealed Fig. S8)12,27. In leptotene spermatocytes, the axial elements complete meiotic arrest in germ cells of Meiob À / À adult males marked by SYCP2 began to form but SYCP1 filaments were not (Fig. 5d,e). In contrast to wild-type seminiferous tubules that yet detectable (Supplementary Fig. S8a). At the zygotene stage, contained a full spectrum of germ cells (Fig. 5d), tubules from axial elements are normally fully formed and chromosomes Meiob À / À testes completely lacked post-meiotic germ cells, with progressively synapse. However, Meiob À / À spermatocytes zygotene-like spermatocytes reflecting the most advanced germ exhibited severe defects in chromosomal synapsis cells (Fig. 5e). In addition, 3-week-old Meiob À / À testis also (Supplementary Fig. S8b–h): 40% of Meiob À / À spermatocytes exhibited a block in meiosis, showing that inaction of Meiob formed long axial elements, suggesting progression to the causes meiotic failure during the first wave of spermatogenesis zygotene stage; however, these cells completely lacked synapsis

NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications 5 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788

MEIOB (1 mM) MEIOB (0.8 mM) MEIOB (mM) EXO1 M ssDNA (nt) M36 8 10 15 18 36 46 60 ssDNA 1.0 2.0 nt (nM)

GST (4 mM) 36 0.1 0.2 0.4 0.6 0.8 0 2 4 1 * Uncut nt *Uncut 15 * Bound 10 * 5 Cleaved 36nt nt 32P-GMP * 10 Free ssDNA 9 8 8 nt 7 * No MEIOB * 7 6 Cleaved 15 Cleaved 10 5

MEIOB (1 mM) EXO1 ssDNA (nt) M365 8101518364660 5′-flap 3′-flap ds (36 bp) ss (60 nt) ss (36 nt) Uncut nt ssDNA * * ssDNA (60 nt) * 15 AMP * * * M * * * MEIOB +++– + 10 MEIOB ++– 5 (1 mM) (1 mM) Uncut Uncut 32P-AMP 32 nt P-ATP 15 10 Cleaved nt 5 No MEIOB 15 32 10 P-AMP 5

Time MEIOB construct 30 bp 34 nt 36 nt 38 nt 46 nt 30 nt (1 mM) * * * * * * * * * * Min 0.1 0.5 1 10 20 30 30 30 30 ′

M RPA2 RPA3 ABCDEFG 3 -flap (nt) 0 4 6816MEIOB (mM) 2 4 8 0.8 0.8 0.8 0.8 0.8 MEIOB – +++++– – – – 0.8 (0.8 mM) * nt * 36 nt nt 36 nt 15 15 >8 nt 10 8 nt 10 5 8 nt 5 <8 nt GMP % Cleaved 0 0 78 15 18 22 42 12 40 Lane 125 34 67 89 10

Figure 4 | MEIOB exhibits ssDNA-specific 30-exonuclease activity. Construct A (MEIOB central region, Fig. 3a) (aa 136–307) was used in experiments illustrated in a through h.(a) Truncated MEIOB exhibits nuclease activity in addition to its ssDNA-binding activity. (b) Size determination of 50-labelled products generated by MEIOB. An 8-nt product was predominant. All products (b through i) were resolved on denaturing acrylamide gels. (c) Truncated MEIOB processes 50-labelled oligonucleotides of various lengths. ExoI digests ssDNA completely. (d) Truncated MEIOB lacks nuclease activity on dsDNA and 50-flap (30 nt) DNA substrates. (e) Truncated MEIOB exhibits 30-exonuclease activity on 30-labelled oligonucleotides of various lengths. EXO1 serves as a positive control. (f) Truncated MEIOB efficiently processes oligonucleotides with a 30-flap (30 nt). (g) Extent of 30-flap digestion by truncated MEIOB. (h) Time-course and gradient analyses of truncated MEIOB nuclease activity. Note that MEIOB at higher concentrations (2 to 8 mM) produced shorter end products (o8 nt). (i) Nuclease activities of various MEIOB mutants shown in Fig. 3a. Recombinant GST-RPA2 and -RPA3 were used as negative controls. Per cent (%) cleavage represents the average from three independent experiments.

(Supplementary Fig. S8b). Such cells were not observed in wild- RAD51/DMC1 together with the HOP2-MND1 heterodimer type males (Supplementary Fig. S8h). In wild-type postnatal day direct strand invasion into the homologous chromosomes, 16 testes, B65% of spermatocytes reached the pachytene resulting in D-loop formation33. RPA may bind to the D-loop, stage. No normal pachytene spermatocytes were observed in whereas ZMM proteins such as MSH4 and TEX11 facilitate the Meiob À / À males (Supplementary Fig. S8h), but B15% of formation of crossovers20,34. These proteins form recombination Meiob À / À spermatocytes appeared to be at a pachytene-like nodules and can be visualized as foci on meiotic chromosomes35. stage, with shortened axial elements, suggesting chromosomal Analysis of chromosome spreads from Meiob À / À spermatocytes condensation, but no evidence for synapsis (Supplementary revealed the presence of the majority of recombination factors Fig. S8f, h). These results demonstrate that MEIOB is essential assessed: gH2AX persisted on chromatin and ATR formed foci on for chromosomal synapsis. synaptonemal complexes (Supplementary Fig. S9). These results are in agreement with the severe synaptic defects in Meiob À / À spermatocytes (Supplementary Fig. S8). The initial numbers of Meiob À / À mice exhibit defects in meiotic recombination.We RAD51 and DMC1 foci between wild-type and Meiob À / À next examined the process of meiotic recombination. During spermatocytes were similar at leptotene (Fig. 6a,b). The meiosis, SPO11-generated DSBs lead to phosphorylation of numbers of RAD51, DMC1, RPA and TEX11 foci were also H2AX through the activation of the ataxia telangiectasia mutated comparable between wild-type and Meiob À / À spermatocytes at kinase28–30. Ataxia telangiectasia and Rad3-related protein (ATR) the zygotene stage (Fig. 6a–d). Interestingly, in the Meiob À / À is recruited to unsynapsed chromosomes and phosphorylates pachytene-like spermatocytes, the number of RAD51 and DMC1 H2AX at the zygotene and pachytene stages31,32. Recombinases foci decreased as expected, but RPA foci persisted (Fig. 6c),

6 NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788 ARTICLE

+/+ –/–

Exon 56 789101112 Wild type 9.0 kb 1.9 kb 2.2 kb 2.2 kb Targeting vector PGKNeo HyTK

Meiob mutant allele PGKNeo 512loxP loxP 11

YBX2 YBX2

P18 testis Adult testes +/– –/– +/+ +/– –/– kDa

MEIOB 50 P0 +/+ –/– 50 ACTB YBX2 YBX2 37

+/+ –/–

RS P1 +/+ –/– Pa ES Zyg-like YBX2 YBX2

+/+ –/–

P2 +/+ –/–

TUNEL TUNEL

TUNEL TUNEL P1 +/+ –/–

Figure 5 | MEIOB is required for meiosis and fertility in mice of both sexes. (a) Targeted inactivation of the Meiob gene. The mouse Meiob gene maps to chromosome 17 and has 14 exons. Deletion of exons 6–10 (aa 111–293) is expected to cause a frame shift in the resulting transcript. (b) Absence of MEIOB protein in P18 Meiob À / À testis. (c) Significant size reduction in 8-week-old Meiob À / À testis. (d,e) Histological analysis of 8-week-old wild-type and Meiob À / À testes. Zyg, zygotene spermatocytes; Pa, pachytene spermatocytes; RS, round spermatids; ES, elongated spermatids. (f,g) TUNEL analysis of wild-type and Meiob À / À seminiferous tubules. Note the large number of apoptotic cells (green) in Meiob À / À tubules. Scale bar, 25 mm(e,g). (h) Histological analysis of ovaries from adult wild-type and Meiob À / À mice. Scale bar, 50 mm. (i–k) Progressive loss of oocytes in Meiob À / À ovaries. Frozen sections prepared from postnatal day P0, P1 and P2 wild-type and Meiob À / À ovaries were immunolabelled with anti-YBX2 antibodies. (l) TUNEL analysis of P1 ovaries. Nuclear DNA (i–l) was counterstained with 4’,6-diamidino-2-phenylindole (DAPI; blue). Scale bars, 50 mm(i–l). suggesting that in the absence of MEIOB, meiotic recombination with full synapsis by 17.5 days of embryonic development, fails after strand invasion, or RAD51/DMC1 filament is unstable whereas 93% of Meiob þ / À oocytes developed to this stage. This and falls off, leading to replacement by RPA1. difference with males may result from the possibility that At the mid-late pachytene stage, MLH1 foci become apparent Meiob À / À oocytes proceed in meiosis with some synapsis in normal spermatocytes, marking sites of crossover36,37. defects, which may be overcome with time, whereas mutant Meiob À / À spermatocytes did not progress to the mid-late spermatocytes with synapsis defects are rapidly eliminated. All pachytene stage and, thus, were devoid of MLH1 foci. The Meiob À / À pachytene oocytes (n ¼ 30) were gH2AX-positive, meiotic defects in fetal oocytes were similar to those observed in indicating that DSBs were not repaired and lacked MLH1 foci spermatocytes from Meiob À / À mice. However, a small percent- (Supplementary Fig. S10). Our results demonstrate that the age of Meiob À / À oocytes (7%) advanced to the pachytene stage markers of early steps of meiotic recombination (DSB formation,

NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications 7 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788

Wild type –/– Meiob SYCP2 RAD51 RAD51 SYCP2 350

300

250 Wild type 200

150

RAD51 foci per cell 100

50 –/–

0 Meiob

=12) =22) =22) =21) =12) =24) =24) =22) n n n n n n n n Le ( Pa ( Le ( e-Zy ( l-Zy ( e-Zy ( l-Zy ( Pa-like (

Wild type Meiob –/– SYCP2 DMC1 DMC1 SYCP2 200

150 Wild type 100 DMC1 foci per cell 50 –/–

0 Meiob =12) =21) =23) =18) =12) =13) =15) =18) n n n n n n n n Le ( Pa ( Le ( e-Zy ( l-Zy ( e-Zy ( l-Zy ( Pa-like (

Wild type Meiob –/– SYCP2 RPA1 RPA1 SYCP2 250

200

150 Wild type

100 RPA foci per cell

50 –/–

0 Meiob =16) =22) =16) =16) =22) =24) =21) =21) n n n n n n n n Le ( Pa ( Le ( e-Zy ( l-Zy ( e-Zy ( l-Zy ( Pa-like (

Wild type Meiob –/– 200 SYCP2 TEX11 TEX11 SYCP2

150 Wild type 100

TEX11 foci per cell 50 –/–

0 Meiob =21) =20) =23) =21) =18) =18) n n n n n n Pa ( e-Zy ( l-Zy ( e-Zy ( l-Zy ( Pa-like (

Figure 6 | MEIOB is essential for meiotic recombination. Immunolabelling for proteins of the synaptonemal complex (SYCP2, lateral elements) and recombination nodules was performed on spread nuclei of spermatocytes from wild-type and Meiob À / À testes at postnatal day 16. The meiotic stages of spermatocytes were determined based on the morphology of the synaptonemal complexes. Spermatocytes were categorized into the following groups: leptotene (Le), early-mid zygotene (e-Zy), late zygotene (l-Zy) and pachytene (Pa). Pachytene-like (Pa-like) spermatocytes contained prominent but short lateral elements (see Supplementary Fig. S8f). Pa-like spermatocytes were present in Meiob À / À testes but absent in wild-type testes. Dot plots for each DNA repair protein show the number of foci per cell; solid lines designate the average for each spermatocyte category; n, number of cells counted. Representative images of spermatocytes at early-mid zygotene stages are shown in separate channels and as merged images. (a) RAD51 foci. (b)DMC1 foci. (c) RPA foci. (d) TEX11 foci. Scale bars, 10 mm.

8 NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788 ARTICLE

50-end resection and strand invasion) are present in the absence indicating that the interaction of RPA with meiotic chromatin is of MEIOB, but that markers of subsequent steps of meiotic independent of MEIOB. In stark contrast, SPATA22 foci were recombination are lacking. Our results suggest that defective absent in Meiob-null spermatocytes and MEIOB foci were absent chromosomal synapsis in Meiob À / À mice is likely to be in Spata22-null spermatocytes (Fig. 7c), suggesting that MEIOB attributed to defects in meiotic recombination. and SPATA22 can interact with meiotic chromosomes only in the presence of each other. Reciprocal interdependence between these two proteins was further supported by the observation that MEIOB forms complexes with RPA2 and SPATA22. To identify SPATA22 protein was absent in Meiob À / À testes and MEIOB protein partners of MEIOB, we performed immunoprecipitation protein was nearly absent in Spata22 À / À testes (Fig. 7d). These (IP) with testicular extracts using anti-MEIOB antibodies and results also indicate that the stability of these two proteins identified a single prominent band that was present in the depends on their coexpression and interaction. Consistent with immunoprecipitated proteins from wild-type but absent from this hypothesis, the Meiob mutant phenocopies the meiotic failure Meiob À / À testes (Fig. 7a). MS/MS analysis revealed that this 38 38 observed in Spata22-null male and female mice . SPATA22 has band contained RPA2 and SPATA22, a meiosis-specific protein . no recognizable domains, whereas MEIOB binds to ssDNA and To verify the association of MEIOB with RPA2 and SPATA22, we exhibits 30-exonuclease activity. These results suggest that MEIOB performed co-IP experiments followed by western blot analysis may function as the catalytic subunit of this novel meiotic and confirmed that RPA2 and SPATA22 were present in the recombination complex that requires SPATA22 as an essential protein fraction immunoprecipitated with the anti-MEIOB cofactor. antibody (Fig. 7b). We found that SPATA22, similar to MEIOB, formed discrete foci on meiotic chromosomes (Fig. 7c and Supplementary Fig. S4b). Specificity of the SPATA22 Discussion antibody was confirmed by the absence of foci in Spata22 À / À Genetic and cytological studies in diverse organisms from yeast to spermatocytes (Supplementary Fig. S4b). In addition, SPATA22 humans have yielded a wealth of knowledge on meiosis. However, co-localized with MEIOB in foci on meiotic chromosomes from more than in other model organisms, approaches to gain insight spermatocytes (Supplementary Fig. S11). These studies into mammalian meiosis have been hampered by various demonstrate that MEIOB forms protein complexes with RPA obstacles. Here we have developed a biochemical and proteomics and with SPATA22 in the testis. strategy to systematically identify meiotic chromatin-associated proteins. Our strategy has been proven to be highly successful, identifying synaptonemal complex components, proteins critical Interdependent localization of MEIOB and SPATA22. RPA for meiotic recombination and factors involved in chromatin foci were formed on Meiob-null meiotic chromosomes (Fig. 6c), modification. In this report, we have characterized the essential

SPATA22 SYCP3 SPATA22 SYCP3 IP: MEIOB

Meiob –/– +/+

kDa Meiob +/+ Meiob –/– 75 MEIOB SYCP3 MEIOB SYCP3 RPA2 25 SPATA22

Spata22 +/+ Spata22 –/–

Meiob Spata22 Extract IP: MEIOB IP: MEIOB Meiob +/+ +/+ –/– +/+–/– +/+ –/– 37 SPATA22 RPA2 37 25 MEIOB 50 SPATA22 * 37 RPA2 25 MEIOB 50 ACTB 37

Figure 7 | MEIOB forms complexes with RPA2 and SPATA22. (a) Identification of MEIOB-associated proteins from P18 testes by IP and mass spectrometry. The gel was stained with SYPRO Ruby. (b) Co-IP analysis of MEIOB with RPA2 and SPATA22 from testicular protein extracts. The asterisk indicates a nonspecific band. (c) Interdependent localization of MEIOB and SPATA22 on meiotic chromosomes from spermatocytes. Spata22 À / À refers to Spata22repro42/repro42 mice38. Scale bar, 10 mm. (d) Western blot analysis of MEIOB and SPATA22 proteins in testes from wild-type, Meiob À / À and Spata22 À / À mutant mice.

NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications 9 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788 function of one novel meiotic chromatin-associated protein— D-loop towards the second end. Subsequent to the second-end MEIOB. Furthermore, these proteomic screen and functional capture, the first end needs to be ligated to the resected strand of studies set the stage for future characterization of other novel the second end. The first end-primed DNA synthesis could putative meiotic chromatin-associated proteins (Supplementary progress beyond the end point of the resected strand of the Table S1). second end, such that a 30-ssDNA flap forms after annealing of Many key steps of meiotic recombination, such as second-end the D-loop with the single-stranded second end or after synthesis- capture, are poorly understood. Biochemical and genetic studies dependent strand annealing (Fig. 8). The timing of DSB end have shown that yeast RAD52 has a key role in second-end resection and first end-primed DNA synthesis is different; thus, capture39–42. In biochemical assays, mammalian RAD52 also generation of a gap or 30-flap is expected after the second-end mediates second-end capture43. However, RAD52-deficient mice annealing. Whereas a gap can be filled by DNA synthesis, the display no abnormalities in viability and fertility, suggesting that repair of a 30-flap requires different mechanisms. MEIOB may RAD52 is not essential for recombination in mice, and that other bind to and remove the 30-flap through its nuclease activity, such genes may functionally substitute RAD52 in mice44. that the ends can be ligated seamlessly without insertions. If Our results support a model in which MEIOB and its cofactor 30-flap is not processed, DNA helicases may act on this SPATA22 may mediate second-end capture during mammalian intermediate and destabilize it. In vivo, RPA may inhibit MEIOB meiotic recombination (Fig. 8). MEIOB is not required for the activity on 30-ends of ssDNA, thereby blocking promiscuous early steps of meiotic recombination, including DSB formation, degradation. It is also possible that binding partners such as RPA end resection and strand invasion. However, crossovers are never and SPATA22 may modulate the nuclease activity and substrate formed in Meiob-deficient germ cells and, in normal pachytene specificity of MEIOB in vivo. spermatocytes, the disappearance of MEIOB foci precedes the In Drosophila and Arabidopsis, RPA1-related proteins appearance of crossovers. These results indicate that MEIOB is (DmHDM and AtRPA1A) have evolved to regulate crossover likely to function at the time point after strand invasion (Fig. 8). formation, but neither is essential for meiotic recombination46,47. Similar to RAD52, MEIOB binds to ssDNA and interacts with Although certain mitotic DNA repair proteins (such as RPA RPA. Therefore, in our model, we propose that MEIOB mediates and RAD51) are utilized during meiotic recombination, many second-end capture through interactions with the RPA proteins, species have developed meiosis-specific repair machinery. One which coat both the D-loop and the resected second end. Second- such factor is DMC1, a meiosis-specific paralogue of RAD51 end capture is a prerequisite to the double Holliday junction (refs 48–50). Yeast meiotic recombination only requires the pathway (Fig. 8)45. The severe synaptic defects and persistence of D-loop forming activity of DMC1, whereas RAD51 has a gH2AX caused by loss of MEIOB suggest that MEIOB is globally supporting role51,52. DMC1 has evolved to initiate and engage required for DSB repair during meiosis. the first interhomologue chromosome interaction unique to We postulate that MEIOB may digest the 30-ssDNA flap from meiosis. Our results suggest that MEIOB might have evolved to the first end after annealing of the second end. After strand drive another interhomologue interaction unique to meiosis— invasion, the first end primes DNA synthesis to extend the capture of the second homologue end.

DSB formation 3′ 5′ 5′ 3′ 5′- to 3′-end resection 3′ 5′ 5′ 3′ 5′ 3′ 3′ 5′ Strand invasion (D-loop) First-end DNA synthesis RPA SPATA22 Second-end 3′ MEIOB 5′ 5′ 3′ 5′ 3′ 3′ 5′ Synthesis-dependent Second-end capture strand annealing (SDSA) Second-end DNA synthesis

3′ 5′ 3′ 5′ 5′ 3′ 5′ 3′ 3′ flap 3′ flap 5′ 3′ 5′ 3′ 3′ 5′ 3′ 5′ dHJ formation

3′ 5′ 3′ 5′ 5′ 3′ 5′ 3′ 5′ 3′ 5′ 3′ 3′ 5′ 3′ 5′ Non-crossover Crossover Figure 8 | Proposed model for MEIOB function during meiotic recombination. Key steps of meiotic recombination are illustrated. We propose that the RPA–MEIOB–SPATA22 complex coats both the D-loop and the ssDNA of the second end. In this model, the interaction between RPA and MEIOB– SPATA22 mediates second-end capture. Most DSBs are repaired through the synthesis-dependent strand annealing (SDSA) pathway, in which the D-loop collapses back to its sister chromatid. After the second-end capture, intermediates continue to form double Holliday junctions (dHJs), whichare resolved into either crossovers or non-crossovers. We propose that MEIOB removes the 30-flaps resulting from the first-end DNA synthesis in both dHJ and SDSA pathways.

10 NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved. NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788 ARTICLE

Methods 350 mgmlÀ 1 G418 and 2 mM ganciclovir and screened by long-distance PCR12. Biochemical purification of chromatin. We optimized previously reported Out of 384 double-resistant embryonic stem cell clones, two homologously targeted chromatin purification protocols (Fig. 1a)10,11 as follows: 100 mg testis tissue was clones were obtained and both yielded germline transmission of the Meiob mutant homogenized in 1 ml Buffer A (250 mM sucrose, 10 mM Tris-HCl, pH 8.0, 10 mM allele through chimeric mice. All offspring were genotyped by PCR. Wild-type 0 MgCl2, 1 mM EGTA, 1  protease inhibitor cocktail III) and centrifuged at 300g to allele (514 bp) was assayed by PCR with primers 5 -AGAGCCTTCTTTCACAA separate nuclei (pellet) from cytoplasmic extract (supernatant). The nuclei were TG-30 and 50-TTTATCTGCCCAGGACTAC-30. The mutant allele (382 bp) was homogenized in Buffer B (Buffer A þ 0.1% Triton X-100, 0.25% NP-40) and assayed by PCR with primers 50-ACCTATACATTGGTGACCTTGA-30 and 50- centrifuged briefly at 100g to remove nuclear membrane particles (pellet), which CCTACCGGTGGATGTGGAATGTGTG-30. Mice were maintained and used for contain some chromatin (discarded). The chromatin-containing supernatant was experimentation according to the guidelines of the Institutional Care and Use layered over a 1.7 M sucrose cushion and centrifuged at 50,000g for 1 h in an Committee of the University of Pennsylvania. ultracentrifuge to remove soluble nuclear proteins and remnant cytoplasmic proteins/organelles. The pellet (chromatin) was washed three times, resuspended in Statistics  . Statistical analysis was performed with Student’s t-test. The values were Tris buffer, sonicated and boiled in an equal volume of 2 SDS–PAGE buffer to ± dissolve all chromatin-associated proteins (B200 mg). presented as mean s.d.

Histological and surface nuclear spread analyses Proteomics screen and mass spectrometry. Seventy-five micrograms of . For histological analysis, chromatin-associated proteins from each testis sample were separated on a 10% testes or ovaries were fixed in Bouin’s solution overnight, embedded with paraffin SDS–PAGE gel (Fig. 1c) and stained with Coomassie blue. The gel lane was cut into and sectioned using a microtome. Sections were processed and stained with ten slices, digested with trypsin and subjected to HPLC followed by MS/MS at the haematoxylin and eosin. For immunofluorescence analysis, testes or ovaries were ° University of Pennsylvania Proteomic Core Facility. All spectra from ten MS/MS fixed in 4% paraformaldehyde for 3 h at 4 C, dehydrated in 30% sucrose, processed runs (one run/slice) were pooled and analysed using the Scaffold 2.6 software. and sectioned. TUNEL assays were carried out with the ApopTag Fluorescein A total of B1,300 proteins were identified from each testis sample. Meiotic In Situ Apoptosis Detection Kit (catalogue nunber S7110, Serologicals Corpora- chromatin-associated proteins were identified by electronic subtraction using tion/Chemicon). For spread analysis, spermatocytes from postnatal day 16 testes Scaffold 2.6 software as follows: meiotic chromatin-associated proteins ¼ P18 were used unless noted otherwise. Testicular tubules or embryonic ovaries were testes À (P6 testes þ XXY* testes; Fig. 1d). extracted in a hypotonic treatment buffer. Cells were suspended in 0.1 M sucrose and were then spread on a thin layer of paraformaldehyde solution containing Triton X-100 (ref. 34). The following primary antibodies were used for Recombinant proteins. All recombinant GST fusion proteins were expressed in immunofluorescence on spread nuclei: SYCP1 (1:50, catalogue number ab15090, Escherichia coli (strain BL21) using the pGEX4T-1 vector. The complementary Abcam), SYCP2 (ref. 12), SYCP3 (1:100, catalogue number 611230, BD DNAs encoding mouse MEIOB, mouse RPA2 and mouse RPA3 were amplified by Biosciences), gH2AX (1:500, catalogue number 16-202A, Clone JBW301, PCR from testis cDNAs and subcloned into the pGEX4T-1 vector. Point mutations Millipore), ATR (1:200, catalogue number PC538, EMD Biosciences), RPA1 (1:50, in MEIOB were generated by PCR-based mutagenesis. All constructs were verified catalogue number ab87272, Abcam), RPA2 (1:100, catalogue number 2208S, clone by sequencing. For induction, 3 ml of Luria broth (LB) medium (100 mgmlÀ 1 4E4, Cell Signaling Technology), RAD51 (1:50, gift from late Peter Moens), DMC1 ampicillin) was inoculated with a single colony, incubated for 6 h at 37 °C and used (1:50, catalogue number sc-22768 (H-100), Santa Cruz Biotech), MLH1 (1:50, to inoculate 50 ml LB medium (100 mgmlÀ 1 ampicillin) at 1:1,000 dilution. After catalogue number 550838, clone G168-15, BD Biosciences), MSH4 (1:50, gift from overnight incubation, 50 ml culture was transferred to 600 ml fresh LB medium Chengtao Her)53, SPATA22 (1:50, catalogue number 16989-1-AP, ProteinTech with ampicillin and continued to incubate at 37 °C till an absorbance at 600 nm of Group) and TEX11 (1:10)34. B1.0. Isopropyl-b-D-thiogalactoside was added to a final concentration of 0.2 mM and the culture was incubated with shaking (180 r.p.m.) at 18 °C for 3 h. Bacteria Co-IP analysis Co-IP was performed with P18 testes using anti-MEIOB anti- were pelleted by centrifugation at 4 °C, resuspended in 20 ml cold DPBS buffer . (Gibco 14190-136) and sonicated on ice. After centrifugation at 12,000g for 20 min bodies (Fig. 7a). Testes (200 mg) were homogenized in 1 ml RIPA buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 1% NP-40, 1 mM dithiothreitol, 0.5% sodium deox- at 4 °C, the supernatant was transferred to a new tube. Three hundred microlitres  of glutathione-Sepharose 4B beads (pretreated with cold PBS) was added to the ycholate, 0.05% SDS, 1 mM EDTA) with 1 protease inhibitor cocktail III. lysate and incubated with continuous mixing for 2 h at 4 °C. The beads were Immunoprecipitated proteins were run on a 4–15% gradient SDS–PAGE gel and stained with SYPRO Ruby (Bio-Rad). The protein band exclusive to the wild-type pelleted by centrifugation for 1 min and washed nine times with 10 ml ice-cold PBS testis sample was subjected to mass spectrometry. each. Proteins were eluted with 1 ml elution buffer (10 mM glutathione, 50 mM For co-IP experiments followed by western blotting (Fig. 7b), 200 mg of P18 Tris-HCl, pH 8.0) for six times and the eluents were pooled. The eluted proteins testes (wild type or Meiob À / À ) were homogenized in 1 ml RIPA buffer plus were concentrated using the Pierce 20K MWCO concentrators (catalogue number   89886A, Thermo Scientific). 1 protease inhibitor cocktail III on ice and 4 ml cold RIPA buffer with 1 protease inhibitor cocktail III, and 90 U benzonase per ml was added to the lysate. The lysate was incubated on a rocking platform at room temperature for 2 h, Electrophoretic mobility shift and nuclease assays. All biochemical assays were followed by centrifugation at 16,100g for 15 min at 4 °C. The supernatant was performed with purified recombinant GST-MEIOB proteins expressed in E. coli. transferred to a new tube and centrifuged again at 100,000g for 1 h at 4 °C. The Nuclease assays were performed for 30 min at room temperature with 1 mM (or supernatant was used for IP with anti-MEIOB antibodies followed by western 0 0 otherwise indicated) recombinant MEIOB protein and 2 nM 5 -or3-end-labelled blotting. Full scans of key western blottings are provided in Supplementary Fig. S12. oligonucleotides (Supplementary Table S2) in binding buffer (5  :50mM Tris- HCl, pH 7.5, 5 mM MgCl2, 2.5 mM EDTA, 2.5 mM dithiothreitol, 250 mM NaCl, 0.25 mg ml À 1 poly(dI-dC):poly(dI-dC), 20% glycerol; Promega). The electro- References phoretic mobility shift binding assays were performed at room temperature for 1. Zickler, D. & Kleckner, N. Meiotic chromosomes: integrating structure and 30 min in the same buffer as for the nuclease assays, except with 10 mM EDTA. function. Annu. Rev. Genet. 33, 603–754 (1999). DNA–protein complexes and DNA products were separated on 10% native or 25% 2. Gerton, J. L. & Hawley, R. S. Homologous chromosome interactions in meiosis: denaturing acrylamide gels, and PhosphorImager radiographs were quantified diversity amidst conservation. Nat. Rev. Genet. 6, 477–487 (2005). using ImageQuant software. 3. Keeney, S., Giroux, C. N. & Kleckner, N. Meiosis-specific DNA double-strand breaks are catalyzed by Spo11, a member of a widely conserved protein family. Cell 88, 375–384 (1997). Generation of polyclonal antibodies. The 6  His-MEIOB (245–466 aa) fusion 4. Mimitou, E. P. & Symington, L. S. DNA end resection: many nucleases make protein was expressed in E. coli using the pQE-30 vector, affinity purified with light work. DNA Repair (Amst) 8, 983–995 (2009). Ni-NTA resin and used to immunize rabbits and guinea pigs (Cocalico Biologicals 5. Szostak, J. W., Orr-Weaver, T. L., Rothstein, R. J. & Stahl, F. W. The double- Inc.), yielding antisera UP2327 (rabbit) and UP-GP 99 (guinea pig). Affinity- strand-break repair model for recombination. Cell 33, 25–35 (1983). purified antibodies were used for immunofluorescence and western blotting 6. Wold, M. S. Replication protein A: a heterotrimeric, single-stranded DNA- analyses. Other antibodies used for western blotting are ACTB (1:7,500, catalogue binding protein required for eukaryotic DNA metabolism. Annu. Rev. Biochem. number A5441, clone AC-15, Sigma), histone H3 (1:2,500, catalogue number 13 66, 61–92 (1997). 05-499, clone 6.6.2, Millipore) and TEX19 (1:500) . 7. Bochkareva, E., Belegu, V., Korolev, S. & Bochkarev, A. Structure of the major single-stranded DNA-binding domain of replication protein A suggests a Targeted inactivation of the Meiob gene. In the targeting construct, a 9-kb dynamic mechanism for DNA binding. EMBO J 20, 612–618 (2001). genomic DNA segment harbouring exons 6–10 of Meiob was replaced with the 8. Soustelle, C., Vedel, M., Kolodner, R. & Nicolas, A. Replication protein A is neomycin selection cassette (Fig. 5a). The two homologous arms (2.2 kb each) were required for meiotic recombination in Saccharomyces cerevisiae. Genetics 161, amplified from a Meiob-containing BAC clone (RP23-397O23) by PCR with high- 535–547 (2002). fidelity DNA polymerase. The thymidine kinase-negative selection marker was 9. Wang, Y. et al. Mutation in Rpa1 results in defective DNA double-strand break inserted adjacent to the right arm. V6.5 embryonic stem cells were electroporated repair, chromosomal instability and cancer in mice. Nat. Genet 37, 750–755 with linearized pUP105/ClaI targeting construct, cultured in the presence of (2005).

NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications 11 & 2013 Macmillan Publishers Limited. All rights reserved. ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms3788

10. Wysocka, J., Reilly, P. T. & Herr, W. Loss of HCF-1-chromatin association 38. La Salle, S. et al. Spata22, a novel vertebrate-specific gene, is required for precedes temperature-induced growth arrest of tsBN67 cells. Mol. Cell. Biol. 21, meiotic progress in mouse germ cells. Biol. Reprod. 86, 45 (2012). 3820–3829 (2001). 39. Mortensen, U. H., Bendixen, C., Sunjevaric, I. & Rothstein, R. DNA strand 11. Chu, D. S. et al. Sperm chromatin proteomics identifies evolutionarily annealing is promoted by the yeast Rad52 protein. Proc. Natl Acad. Sci. USA 93, conserved fertility factors. Nature 443, 101–105 (2006). 10729–10734 (1996). 12. Yang, F. et al. Mouse SYCP2 is required for synaptonemal complex assembly 40. Sugiyama, T., Kantake, N., Wu, Y. & Kowalczykowski, S. C. Rad52-mediated and chromosomal synapsis during male meiosis. J. Cell Biol. 173, 497–507 DNA annealing after Rad51-mediated DNA strand exchange promotes second (2006). ssDNA capture. EMBO J. 25, 5539–5548 (2006). 13. Yang, F. et al. The ubiquitin ligase Ubr2, a recognition E3 component of the 41. Lao, J. P., Oh, S. D., Shinohara, M., Shinohara, A. & Hunter, N. Rad52 N-end rule pathway, stabilizes Tex19.1 during spermatogenesis. PLoS One 5, promotes postinvasion steps of meiotic double-strand-break repair. Mol. Cell e14017 (2010). 29, 517–524 (2008). 14. McCarrey, J. R. In Cell and Molecular Biology of the Testis. (eds Desjardins, C. & 42. Nimonkar, A. V., Sica, R. A. & Kowalczykowski, S. C. Rad52 promotes second- Ewing, L.) 58–89 (Oxford University Press, 1993). end DNA capture in double-stranded break repair to form complement- 15. Hunt, P. A. & Eicher, E. M. Fertile male mice with three sex chromosomes: stabilized joint molecules. Proc. Natl Acad. Sci. USA 106, 3077–3082 (2009). evidence that infertility in XYY male mice is an effect of two Y chromosomes. 43. McIlwraith, M. J. & West, S. C. DNA repair synthesis facilitates RAD52- Chromosoma 100, 293–299 (1991). mediated second-end capture during DSB repair. Mol. Cell 29, 510–516 (2008). 16. Plug, A. W. et al. Changes in protein composition of meiotic nodules during 44. Rijkers, T. et al. Targeted inactivation of mouse RAD52 reduces homologous mammalian meiosis. J. Cell. Sci 111 (Pt 4), 413–423 (1998). recombination but not resistance to ionizing radiation. Mol. Cell. Biol. 18, 17. Moens, P. B. et al. The time course and chromosomal localization of 6423–6429 (1998). recombination-related proteins at meiosis in the mouse are compatible with 45. Svendsen, J. M. & Harper, J. W. GEN1/Yen1 and the SLX4 complex: Solutions models that can resolve the early DNA-DNA interactions without reciprocal to the problem of Holliday junction resolution. Genes Dev. 24, 521–536 (2010). recombination. J. Cell. Sci. 115, 1611–1622 (2002). 46. Joyce, E. F., Tanneti, S. N. & McKim, K. S. Drosophila hold’em is required for a 18. Moens, P. B., Marcon, E., Shore, J. S., Kochakpour, N. & Spyropoulos, B. subset of meiotic crossovers and interacts with the dna repair endonuclease Initiation and resolution of interhomolog connections: crossover and non- complex subunits MEI-9 and ERCC1. Genetics 181, 335–340 (2009). crossover sites along mouse synaptonemal complexes. J. Cell. Sci. 120, 47. Osman, K., Sanchez-Moran, E., Mann, S. C., Jones, G. H. & Franklin, F. C. 1017–1027 (2007). Replication protein A (AtRPA1a) is required for class I crossover formation but 19. Tarsounas, M., Pearlman, R. E. & Moens, P. B. Meiotic activation of rat is dispensable for meiotic DNA break repair. EMBO J. 28, 394–404 (2009). pachytene spermatocytes with okadaic acid: the behaviour of synaptonemal 48. Bishop, D. K., Park, D., Xu, L. & Kleckner, N. DMC1: a meiosis-specific yeast complex components SYN1/SCP1 and COR1/SCP3. J. Cell. Sci. 112 (Pt 4), homolog of E. coli recA required for recombination, synaptonemal complex 423–434 (1999). formation, and cell cycle progression. Cell 69, 439–456 (1992). 20. Kneitz, B. et al. MutS homolog 4 localization to meiotic chromosomes is 49. Yoshida, K. et al. The mouse RecA-like gene Dmc1 is required for homologous required for chromosome pairing during meiosis in male and female mice. chromosome synapsis during meiosis. Mol. Cell 1, 707–718 (1998). Genes Dev. 14, 1085–1097 (2000). 50. Pittman, D. L. et al. Meiotic prophase arrest with failure of chromosome 21. Flynn, R. L. & Zou, L. Oligonucleotide/oligosaccharide-binding fold proteins: a synapsis in mice deficient for Dmc1, a germline-specific RecA homolog. growing family of genome guardians. Crit. Rev. Biochem. Mol. Biol. 45, 266–275 Mol. Cell 1, 697–705 (1998). (2010). 51. Cloud, V., Chan, Y. L., Grubb, J., Budke, B. & Bishop, D. K. Rad51 is an 22. Theobald, D. L., Mitton-Fry, R. M. & Wuttke, D. S. Nucleic acid recognition by accessory factor for Dmc1-mediated joint molecule formation during meiosis. OB-fold proteins. Annu. Rev. Biophys. Biomol. Struct. 32, 115–133 (2003). Science 337, 1222–1225 (2012). 23. Bochkarev, A., Pfuetzner, R. A., Edwards, A. M. & Frappier, L. Structure of the 52. Bishop, D. K. Rad51, the lead in mitotic recombinational DNA repair, plays a single-stranded-DNA-binding domain of replication protein A bound to DNA. supporting role in budding yeast meiosis. Cell. Cycle 11, 4105–4106 (2012). Nature 385, 176–181 (1997). 53. Her, C., Wu, X., Griswold, M. D. & Zhou, F. Human MutS homologue MSH4 24. Roeder, G. S. & Bailis, J. M. The pachytene checkpoint. Trends Genet. 16, physically interacts with von Hippel-Lindau tumor suppressor-binding protein 395–403 (2000). 1. Cancer. Res. 63, 865–872 (2003). 25. Yang, J. et al. Absence of the DNA-/RNA-binding protein MSY2 results in male and female infertility. Proc. Natl Acad. Sci. USA 102, 5755–5760 (2005). 26. Di Giacomo, M. et al. Distinct DNA-damage-dependent and -independent Acknowledgements responses drive the loss of oocytes in recombination-defective mouse mutants. We thank B. Kaufman for advice on biochemical purification, C.X. Yuan for help with Proc. Natl Acad. Sci. USA 102, 737–742 (2005). mass spectrometry, R. Schultz for the YBX2 antibody, late P. Moens for the RAD51 27. de Vries, F. A. et al. Mouse Sycp1 functions in synaptonemal complex antibody, C. Her for the MSH4 antibody, J. Kolesar for help with ImageQuant analysis, assembly, meiotic recombination, and XY body formation. Genes Dev. 19, Z. Deng and S. Kasowitz for help with sequence alignments and S. Eckardt for help with 1376–1389 (2005). manuscript preparation. NIH grants GM089893 and GM076327 supported P.J.W.; 28. Fernandez-Capetillo, O., Liebe, B., Scherthan, H. & Nussenzweig, A. H2AX HD42137 supported M.A.H.; funds from Midwestern University supported S.L.; and the regulates meiotic telomere clustering. J. Cell Biol. 163, 15–20 (2003). Priority Program 1384 of the Deutsche Forschungsgemeinschaft supported R.B. 29. Bellani, M. A., Romanienko, P. J., Cairatti, D. A. & Camerini-Otero, R. D. SPO11 is required for sex-body formation, and Spo11 heterozygosity rescues Author contributions the prophase arrest of Atm-/- spermatocytes. J. Cell. Sci. 118, 3233–3245 M.L. and F.Y. carried out most experiments and data analysis; N.A.L. performed blas- (2005). tocyst injection and generated chimeric mice; J.L., S.L. and M.A.H. contributed with 30. Lange, J. et al. ATM controls meiotic double-strand-break formation. Nature SPATA22-related experiments; R.B. performed the electron microscopy experiment; 479, 237–240 (2011). P.J.W. conceptualized the proteomics screen, analysed data and wrote the manuscript. 31. Turner, J. M. et al. Silencing of unsynapsed meiotic chromosomes in the mouse. All the authors were involved in the discussion and commented on the manuscript. Nat. Genet. 37, 41–47 (2005). 32. Burgoyne, P. S., Mahadevaiah, S. K. & Turner, J. M. The consequences of asynapsis for mammalian meiosis. Nat. Rev. Genet. 10, 207–216 (2009). Additional information 33. Petukhova, G. V. et al. The Hop2 and Mnd1 proteins act in concert with Rad51 Supplementary Information accompanies this paper at http://www.nature.com/ and Dmc1 in meiotic recombination. Nat. Struct. Mol. Biol. 12, 449–453 (2005). naturecommunications 34. Yang, F. et al. Meiotic failure in male mice lacking an X-linked factor. Genes Dev. 22, 682–691 (2008). Competing financial interests: The authors declare no competing financial interests. 35. Handel, M. A. & Schimenti, J. C. Genetics of mammalian meiosis: regulation, Reprints and permission information is available online at http://npg.nature.com/ dynamics and impact on fertility. Nat. Rev. Genet. 11, 124–136 (2010). reprintsandpermissions/ 36. Edelmann, W. et al. Meiotic pachytene arrest in MLH1-deficient mice. Cell 85, 1125–1134 (1996). How to cite this article: Luo, M. et al. MEIOB exhibits single-stranded DNA-binding 37. Baker, S. M. et al. Involvement of mouse Mlh1 in DNA mismatch repair and and exonuclease activities and is essential for meiotic recombination. Nat. Commun. meiotic crossing over. Nat. Genet. 13, 336–342 (1996). 4:2788 doi: 10.1038/ncomms3788 (2013).

12 NATURE COMMUNICATIONS | 4:2788 | DOI: 10.1038/ncomms3788 | www.nature.com/naturecommunications & 2013 Macmillan Publishers Limited. All rights reserved.